In the mobile Internet era, people usually take a screenshot on a mobile phone for sharing. However, when content of more than one screen needs to be shared, screenshots need to be consecutively taken for content in a plurality of screens. Operations are relatively complex, and readability is poor. Therefore, applications such as “long screenshot” and “scrolling screenshot” appear. Key experience of the long screenshot application or the scrolling screenshot application is whether a final stitched “long image” is accurate. However, for whether the long image can be stitched accurately, a key technology is to accurately identify an overlapped area between to-be-stitched screenshots, and remove the overlapped area for stitching.
An image comparison method is usually used in the prior art. Whether there is an overlapped area between the “bottom” of a screenshot (an (N−1)th screenshot) and the “top” of a next screenshot (an Nth screenshot) is analyzed pixel by pixel. According to the existing image comparison method, for a scenario using a solid-color chat background, an overlapped area between the two images can be accurately identified. However, once the user uses a non-solid-color chat background, a relative location of the “chat background” always changes, and the overlapped area between the two images cannot be found by using only the image comparison method. As a result, the stitched image has overlapped content, and accuracy of similar image stitching needs to be further improved.